carlo a. favero* and alessandro missale contagion in the ... · carlo a. favero* and alessandro...

Carlo A. Favero* and Alessandro Missale

Contagion in the EMU – The Roleof Eurobonds with OMTs

DOI 10.1515/rle-2016-0043Published online October 7, 2016

Abstract: We find strong evidence of country interdependence in the pricing ofdefault risk, which suggests that a crisis can easily propagate from countries withweak fiscal fundamentals to other fiscally sounder member States. Interest rateinterdependence differs between countries with high interest rates – high yielders– and countries with low interest rates – low yielders –. The former countries arelinked through global spreads; i. e. they are exposed to the interest rate spreads(over Germany) of other troubled countries to a degree which increases with fiscalproximity. Low yielders with sounder fiscal fundamentals are partially immunefrom the high interest rates of fiscally weak member States but are still exposed tothe risk of a euro break-up that is priced in Quanto CDS. This “euro risk” factor is amain driver of the interest rate spreads of low yielders until August 2012. Moreimportantly, our case study of Italy shows that the impact of the global spreadvariable is dominated by changes in market sentiment, a sign that the Italian2011–2012 crisis had the characteristics of a debt run more than a crisis offundamentals. This evidence suggests that Eurobonds would be justified as aninstrument for crisis prevention in the absence of a “lender of last resort”. With theannouncement of OMTs, the ECB seems to have taken such role upon itself,mainly as a response to the risk of a euro break-up. We show that OMTs led toa significant fall in the impact effect of the global spread variable in the Italiancase. The ECB’s ability to buy member States’ bonds reduces the risk of a self-fulfilling debt run but also deprives Eurobonds of their role in crisis prevention.Proposals to introduce Eurobonds to finance investment projects and expendituresrelated to the security and refugee crisis appear more realistic.

Keywords: Eurobonds, global VAR, bond spreads in the euro-area, defaultpremium, liquidity premium, Euro break-up, ESM, OMTsJEL Classification: C51, C58

*Corresponding author: Carlo A. Favero, Deutsche Bank Chair in Asset Pricing and QuantitativeFinance, Department of Finance, Bocconi University, IGIER and CEPR, Milano, Italy,E-mail: [email protected] Missale, Department of Economics, Management and Quantitative Methods(DEMM), Università degli Studi di Milano, Milano, Italy, E-mail: [email protected]

Rev. Law Econ. 2016; 12(3): 555–584

Brought to you by | Università degli Studi di MilanoAuthenticated

Download Date | 5/4/17 12:43 PM

1 Introduction

After reaching a peak in November 2011 and again in the spring of 2012, interestrate differentials, i. e. yield spreads, between government bonds in the euro areahave finally narrowed back to pre-crisis levels, marking the end of the Europeansovereign debt crisis. The ESM–ECB framework for financial assistance has beensuccessful in stabilizing investors’ expectations and restoring market confi-dence. In particular, the ECB announcement of Outright Monetary Transactions(OMTs) in August 2012 seems to have played a key role in reducing the Italianand Spanish yield spreads over German Bunds.

Eurobond proposals during the European debt crisis of 2010–2012 focusedon their potential role in preventing a self-fulfilling liquidity crisis not justifiedby fiscal fundamentals and its spreading to the whole euro area. These propo-sals were based on the idea that Eurobonds jointly guaranteed by all memberStates would ensure continuous market access to countries facing a liquiditycrisis. Although it is always difficult to distinguish whether a crisis is one ofsolvency or liquidity, as both aspects are usually present at the same time, inthis paper we report evidence that, starting in the second half of 2011, the Italianyield spread over Germany was significantly affected by changes in marketsentiment. We extend Favero and Missale (2012) to consider a common “eurorisk” factor reflecting euro devaluation risk, and possibly the risk of a eurobreak-up, that we measure through the average Quanto CDS of euro-area coun-tries. We focus on the period of the European sovereign debt crisis fromNovember 2009 to November 2015 and split this sample into two sub-periods,before and after August 2012, in order to assess the impact of the ECB announce-ment of OMTs.

We find strong evidence of country interdependence in the pricing ofdefault risk, which suggests that a crisis can easily propagate from countrieswith weak fiscal fundamentals to other fiscally sounder member States.However, interest rate interdependence differs between countries with highinterest rates – high yielders – and countries with low interest rates – lowyielders-, and over time. The former countries are linked through globalspreads; i. e. they are exposed to the yield spreads of other troubled countriesto a degree which increases with fiscal proximity. Low yielders with sounderfiscal fundamentals are partially immune from the high yield spreads offiscally weak member States but are still exposed to the risk of a devaluationor break-up of the euro that is priced in Quanto CDS. This “euro risk” factor isthe main driver of the yield spreads of low yielders until August 2012, whichsuggests that a broader crisis of the euro area is the channel through which

556 C.A. Favero and A. Missale



a debt default in a country with weak fundamentals can propagate to safermember States. More importantly, our case study of Italy shows that the impactof the global spread variable is dominated by changes in market sentiment –a sign that the Italian 2011–2012 crisis had the characteristics of a debt runmore than a crisis of fundamentals. This evidence suggests that the introduc-tion of Eurobonds as an instrument for crisis prevention would be economic-ally justified. Eurobonds would be in the interest not only of States withfinancing difficulties but also of States with stronger fundamentals in that adebt default of, say, Italy or Spain would rapidly propagate to the wholeMonetary Union.

The other rationale for Eurobonds is the absence of a “lender of last resort”who can prevent a debt run.1 With the announcement of OMTs, the ECB seems tohave taken such role upon itself, mainly as a response to the risk of a eurobreak-up. This has led to a steady decline in Quanto CDSs and to a significantfall in the yield spreads of Italy and Spain over Germany. The stabilizing effectthat the ECB announcement had on market expectations is well evidenced in oureconometric analysis by the reduction in the impact effect of the global spreadvariable in the Italian case. As the ECB’s decision to buy government bonds ofmember States receiving financial assistance from the ESM reduces the risk of aself-fulfilling debt run, it also weakens the case for using Eurobonds as a crisisprevention instrument.2

The ESM–ECB framework for crisis prevention has stabilized investors’expectations and, most likely, eliminated the need for Eurobonds even if theECB cannot act as a genuine lender of last resort.3 More importantly, theintroduction of Eurobonds for crisis prevention faces strong political oppositionbecause it raises moral hazard issues and entails economic costs that safermember States are not willing to pay. By contrast, proposals to issue

1 The ECB cannot act as a genuine lender of last resort as Art. 123(1) TFEU prohibits the ECBfrom providing any type of credit facility to governments or other public authorities, and fromdirectly purchasing their debt instruments. In fact, OMTs have been justified as secondarymarket bond purchases in the context of the ECB’s open market operations (Art. 18 of‘Protocol No.4 on the Statute of the European System of Central Banks and the EuropeanCentral Bank) with the aim of “safeguarding an appropriate monetary policy transmissionand the singleness of the monetary policy” (ECB, 2012).2 The fulfillment of this expectation also hinges on compliance with Art. 123(1) TFEU and thejurisprudence of national constitutional courts.3 The ECB is not allowed to buy government bonds directly from member States (Art. 123(1)TFEU), and “a necessary condition for OMTs is strict and effective conditionality attached to anappropriate ESM programme” (ECB, 2012).

Contagion in the EMU 557



Eurobonds for expenditures that provide clear benefits to all member States maygain political consensus. This would be the case of investment projects and ofextraordinary expenditures to deal with security concerns and the refugee crisis.As these problems are common to all member States and the expendituresincurred by each member State have clear positive externalities, Eurobondsseem the appropriate instruments to share the cost and the credit risk associatedwith their financing.

The rest of this paper is organized as follows. After this Introduction,Section 2 discusses the potential roles and proposals of Eurobonds. In Section3 we report a graphical decomposition of interest rate spreads between euro-areagovernment bonds and German bonds into their default and non-default com-ponents, and we explore their behavior and determinants. Section 3 presents theestimation results of the Global VAR model of euro-area sovereign spreadsintroduced by Favero and Missale (2012) and Favero (2013). This section alsoaddresses the issue of contagion using a multi GARCH model, and reports anestimate of the effect of the ECB announcement of OMTs on the Italian yieldspread. Section 5 discusses the implications for the role of Eurobonds of OMTsand, more generally, of the ESM–ECB scheme of financial assistance. Section 6concludes the analysis with a proposal for introducing Eurobonds.

2 What use for Eurobonds?

Considering alternative objectives that Eurobonds backed by (several and) jointguarantees can help to achieve is a better starting point than stating theirbenefits and costs because, depending on these objectives, Eurobonds shouldhave different characteristics, be issued in different amounts, and possibly willhave different costs.

Eurobonds can be used to achieve different policy goals. In this paper wefocus on three main objectives: (i) to promote further integration of euro areabond markets; (ii) to halt a debt crisis not justified by fundamentals, i. e. a self-fulfilling “debt run”; (iii) to finance EU-wide projects such as infrastructure orextraordinary spending needs. We review each of them in turn.

2.1 Bond market integration

Promoting further integration of euro area financial markets (perhaps, togetherwith greater fiscal coordination) was the “classic” objective in the original




proposal of Eurobonds by the Giovannini Group Report (2000), later followed bythe Monti Report (2010). Apart from lower segmentation and better functioningof euro financial markets, the proposal hinges on the idea that a fully integratedmarket would deliver liquidity gains to all and, possibly, boost the role of theeuro as a reserve currency. In particular, small issuers (representing a tinyfraction of the euro bond market) paying a high liquidity premium would mostlybenefit from a jointly guaranteed Eurobond, but even Germany would benefitfrom having its bonds traded in a market potentially as thick and liquid as theUS market. Evidence of a sizeable, say, 40 basis-point liquidity premium in favorof US Treasuries over German Bunds was taken as the potential reduction inborrowing costs that would derive from greater liquidity. Summing up, thisargument for Eurobonds focuses on secondary markets and its relevance isstrictly related to the importance of liquidity premia in bond yields. Two impor-tant points must be made at this stage. First, since a large market size is anecessary (but not sufficient) condition to reap liquidity gains, the outstandingamount of Eurobonds should be large, say, comparable to US Treasuries. Thiscannot be done overnight, unless through redenomination, which howeverraises a number of economic and legal issues. The second important point isthat, unlike crisis prevention, market integration offers a rationale for Germanparticipation. As credit risk mutualisation penalizes safe borrowers, liquidity isthe only reason why Germany might reduce its borrowing costs and directlybenefit from the issuance of Eurobonds.

2.2 Debt crisis prevention

The objective to avoid a debt crisis not justified by fiscal fundamentals and itspropagation to safer member States has been at the center stage of Eurobondproposals during the European sovereign debt crisis of 2010–2012. The idea isthat Eurobonds guaranteed by all member States would ensure continuousmarket access for countries with weaker fiscal fundamentals and thus moreexposed to changes in market sentiment. The possibility to rely on Eurobondswould be especially valuable to countries hit by a roll-over crisis when thedemand for new bonds dries up because of each investor’s fears that otherinvestors will shy away in a panic equilibrium reminiscent of a self-fullingbank run. Absent a central bank acting as a lender of last resort, a self-fulfillingdebt run is possible and Eurobonds may provide an alternative to central bankintervention in stabilizing investors’ expectations and preventing a panic equili-brium. To the extent that a debt crisis widens interest rate spread on Bunds,impairs borrowing conditions in the euro area, and propagates to safe member




States, Eurobonds may clearly benefit more countries than just those experien-cing a debt run. The insurance that Eurobonds would provide to States withweaker fundamentals, like Italy and Spain, would also work as insurance forothers. Although it is always difficult to distinguish between solvency andliquidity crises, as solvency is inherently difficult to evaluate and both aspectsare usually present at the same time, Favero and Missale (2012) report evidenceon Italian and Spanish interest rate spreads over Germany that points to asignificant component of such spreads due to changes in market sentimentsince the summer of 2011.

Interestingly, crisis prevention arguments focus on the primary marketand are not concerned with achieving a large market size. In principle,Eurobonds might not even need to be issued since what matters (as in thecase of OMTs) is the reassuring effect on investors’ expectations of endowingcountries with an instrument for market access of last resort. However, it isworth noting that risk mutualisation would not be without costs for safermember States; even in the case that Eurobonds would not be issued by acountry with financing difficulties, the option to do it would increase theexpected liabilities of other member States. For this reason, Eurobonds jointlyguaranteed by all member States may possibly raise the borrowing costs onall type of bonds.

The establishment of the European Stability Mechanism (ESM) and theESM–ECB framework for financial assistance have likely deprived Eurobondsof their role in crisis prevention, in that a country experiencing a liquidity crisiscan now rely on ESM financial assistance, i. e. a credit line to be used for bondpurchases on the primary market, and on OMTs, i. e. ECB bond purchases on thesecondary market.

2.3 EU-wide project financing

The third argument for Eurobonds is that they are the appropriate instruments tofinance projects and/or expenditures that provide benefits to all member States(in proportions that cannot be assessed) also because of spillover effects and/orexternalities. This is clearly the case for infrastructure such as roads and rail-ways connecting two or more member States that facilitate mobility within theEU. However, a similar argument can be made for any investment project that,independently of location, increases output across the EU through spillovereffects. More recently, it has been argued that extraordinary expenditures todeal with the refugee crisis and expenditures for security following terroristattacks should be financed with Eurobonds. As these problems are common to




all member States and the expenditures incurred by each member State haveclear positive externalities to other States, sharing the cost and the credit riskassociated with their financing is economically justified. A potential advantageof Eurobond financing would be to create fiscal space in national budgetswithout lengthy negotiations on budget flexibility.

It is worth noting that a Eurobond program for specific projects would belimited in the amount of bonds outstanding so that a liquidity premium willhave to be paid. Nevertheless, the introduction of Eurobonds backed by (severaland) joint guarantees would represent an important first experiment to testinvestors’ demand and borrowing costs. In fact, bonds issued by the ESM orthe European Investment Bank are guaranteed by the capital that member Stateshave provided to such institutions; they are, therefore, similar to bonds backedby several but not joint guarantees. Finally, the announcement of a commonprogram would enhance the credibility of EU institutions by signaling a politicalwill for greater fiscal unity and cooperation, thus paving the way for a deeperreform of EU fiscal governance.

3 Default versus non-default component

In this section we examine evidence on long-term interest rate differentialsbetween euro-area government bonds and German bonds for the period 2007–2015 in order to identify their determinants. In particular, we aim to disen-tangle interest differentials into their main components: a default premiumand a non-default premium reflecting liquidity risk and factors related toexpected variations in the exchange rate.4 Assessing the relative importanceof such components is crucial to understand the role that Eurobondscan play.

Long-term interest rate differentials between 10-year euro-area governmentbonds and German Bunds co-move with an unstable pattern over time. Interestspreads over Germany converged significantly with the introduction of the euro,narrowing from highs in excess of 300 basis points in the pre-EMU period toless than 30 basis points about one year after the introduction of the euro. Yet,bonds issued by euro-area member States have never been regarded as perfectsubstitutes by market participants. Interest rate differentials co-moved synchro-

4 Bernoth et al. (2012), Di Cesare et al. (2012) and Calvori et al. (2016) also consider an “exces-sive fear” related component.




nously at a very low level between the start of the EMU and the subprime loancrisis. Differentials became sizeable during 2008 and 2009 with some separa-tion in co-movement between high-debt and low-debt countries. The Europeansovereign debt crisis from the end of 2009 onwards produced interest differ-entials of the same or even greater magnitude than those of the pre-EMU era,and more heterogeneity in co-movement. Interest rates have progressivelyreturned to a convergence pattern after the announcement by the ECB ofOMTs in August 2012.

3.1 Default premium, liquidity risk and devaluation risk

Interest rate differentials, i. e. yield spreads, between government bonds of euro-area countries should price three factors: default risk, liquidity risk and expecteddevaluation, say, because of currency redenomination. The latter factor mayemerge, for example, because of a euro break-up following a major credit eventor a German exit from the Eurozone.

Sovereign issuers that are perceived as having a greater solvency risk mustpay investors a default premium. Liquidity risk is the risk of having to sell (orbuy) a bond in a thin market and, thus, at a loss of value and/or highertransaction costs. Small issuers with low volumes of bonds outstanding andthus small markets must compensate investors with a liquidity premium. Theintroduction of the euro in January 1999 initially eliminated the expectationsof exchange rate variations, but the general surge in the debt-to-GDP ratio ofeuro-area countries after 2009 has led markets to reconsider the possibility ofsome member States’ exit from the euro and even the collapse of the commoncurrency.

Favero and Missale (2010, 2012) show that default risk is the main driver ofyield spreads, with non-default risk playing a role only in the case of smallissuers and, more generally, during the global financial crisis. The non-defaultpremium, indeed, appears to be very small with the exception of Finland,the Netherlands, and France in 2008 and 2009. This component is clearlytime-varying and fluctuates between around 10 basis points in calm periodsand around 50 basis points during crises. The risk of devaluation of euro-denominated assets in case of a major credit event, possibly leading to abreak-up of the euro, is not considered in Favero and Missale (2012) as theirsample period ends in August 2011, when such risk only began to be perceived.In what follows, we extend the analysis to consider the impact of a proxy foreuro devaluation risk.




3.2 Default, liquidity and devaluation risk: the evidence

We extend the original framework in Favero and Missale (2012) by keeping thesame measure of default risk and by introducing a proxy for euro devaluationrisk. We use data associated with 10-year benchmark government bonds issuedby ten European countries, namely, Austria, Belgium, Finland, France, Germany,Ireland, Italy, Portugal, Spain, and the Netherlands which are the most fre-quently traded bonds in local and Euro MTS markets. Bond yields, i. e. interestrates, are taken from Datastream.

Figure 1 graphs the 10-year maturity bond yields of the countries underanalysis. The graph suggests that the European sovereign bond market hasexperienced three different phases. Until to 2009, bond yields tend to co-movestrongly and their variation is fairly low. From the start of the European sover-eign debt crisis in late 2009, the bond yields start to diverge. The gap betweenyields across member States increased throughout 2011. Beginning in 2012, theEuropean sovereign bond market showed clear signs of segmentation. Bondscould be classified into high yielders (Ireland, Italy, Spain and Portugal) and lowyielders (Germany, Austria, Belgium, Finland, France, and the Netherlands).This segmentation is very important in explaining the dynamics of yield spreadswith respect to their driving factors.

0

10

20

30

40

50

2008 2009 2010 2011 2012 2013 2014 2015

GermanyBelgiumSpainFinlandGreeceIrelandItalyThe NetherlandsAustriaPortugal

10Y Yields

Figure 1: 10-Y Government bond yield in the Euro area.Note: Yields in % annual termsSource: Datastream/Thomson Financial.




Default risk is measured using CDS spreads. A CDS is a swap contract in whichthe protection buyer of the CDS makes a series of premium payments to theprotection seller and, in exchange, receives a payoff if the bond goes intodefault. The difference between a CDS on a member State bond and the CDSon the German Bund of the same maturity is a measure of the default riskpremium of that State relative to Germany.

To capture the risk of devaluation of euro-denominated assets, which ispossibly associated with the risk of a euro break-up, we use the Quanto CDS, ameasure introduced by De Santis (2015). The Quanto CDS is defined as thedifference between the CDS quotes in US dollar and euros for the samecountry. That difference between the US dollar – and the euro-denominatedCDS spreads represents the premium that market participants are willing topay to receive protection in US dollars so as to avoid the risk that the euro (oran eventual new legacy currency) depreciates against the dollar after a defaultin a member State. De Santis shows that the Quanto CDS is equal to theexpected (present value) loss from receiving protection in euros instead ofdollars in case of default, and it thus depends on the probabilities of defaultin future periods of the contract multiplied by the payout upon default and theexpected rate of depreciation in these future periods. In other words, theQuanto CDS is a “credit-event-probability-weighted” currency premium asso-ciated with the payout from default. This measure is available from 2010onwards. Figure 2 graphs 10-year maturity Quanto CDS rates for the countriesunder analysis.

The behavior of the Quanto CDS of euro-area countries is characterized bya strong co-movement along a common trend with cross-country differencesthat vary within a 50 basis point band (with Spain exhibiting wider fluctua-tions). In fact, Quanto CDS differ across euro-area countries because of differ-ent probabilities of default and payouts, but they are characterized by acommon trend in that a credit event in one country will affect the QuantoCDS of other countries through expected depreciation, currency risk anddefault probabilities. Therefore, the common trend displayed in Figure 2 repre-sents the euro devaluation risk that, we think, correlates with the risk of a eurobreak-up.5 Having measured the Quanto CDS for each country, we take thesimple average across all euro-area countries to eliminate the effect of country-specific factors and outliers. The average Quanto CDS constitutes a common

5 It is worth noting that a credit event does not imply the break-up of the euro area, as in theGreek case in March 2012. Quoting De Santis (2015), “the Quanto CDS measures the riskassociated with the depreciation against the US dollar and not the intra-euro area currencyrisk”.




risk factor that captures expectations of underperformance of the Euro withrespect to the US dollar upon occurrence of a credit event that triggers CDSpayments. We call this common factor the “devaluation risk” or “euro risk”factor.

Figure 3(a) and (b) report the yield spreads for euro-area countries relative toGermany, along with their default and non-default components. We group euro-area countries into low yielders in Figure 3(a) and high yielders in Figure 3(b).

–.010

–.005

.000

.005

.010

.015

.020

.025

.030

I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV

2010 2011 2012 2013 2014 2015

GermanyBelgiumSpainIrelandItalyThe NetherlandsAustriaPortugalAverage

Quanto spreads

–.002

.000

.002

.004

.006

.008

.010

.012

I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV

2010 2011 2012 2013 2014 2015

GermanyItaly

Quanto spreads

Figure 2: Quanto spreads.Source: Datastream/Thomson Financial.




In particular, we report the yield spreads between 10-year government bondsand German Bunds (blue line) along with: the associated CDS spreads (red line),the residual non-default component (black line), and the “euro risk” factor, i. e.the average Quanto CDS in the euro area (green line).The following facts emerge from the data:

i. The first important fact about the co-movements of yield spreads in theeuro area is that their interdependence is not constant over time, and itdiffers between countries with high interest rates – high yielders – andcountries with low interest rates – low yielders-.

ii. In the case of high yielders, the CDS differential relative to Germany, i. e.the default risk component of the yield spread, accounts for virtually theentire yield differential (and its variability) from the beginning of thesample to the end of 2010. The gap between the yield spread and itsdefault component that emerges between 2010 and 2012 is strongly relatedto the “euro risk” factor. With the steady decline in the risk of a euro

–.008

–.004

.000

.004

.008

.012

2008 2009 2010 2011 2012 2013 2014 2015

Yields Componenents FN

–.004

.000

.004

.008

.012

.016

.020

2008 2009 2010 2011 2012 2013 2014 2015

Yields Componenents FR

–.002

.000

.002

.004

.006

.008

.010

2008 2009 2010 2011 2012 2013 2014 2015

Yields Componenents NL

–.010

–.005

.000

.005

.010

.015

.020

2008 2009 2010 2011 2012 2013 2014 2015

yield spread vs GERcds spread vs GERnon default component vs GERredenomination risk

Yields Componenents OE

Figure 3a: The default and non-default component in yields spreads – Low Yielders.Source: Datastream/Thomson Financial.




break-up, from August 2012 onward, the non-default component goesback to fluctuating close to zero.

iii. The case of low-yielders is mixed. In Austria and France the default riskcomponent of the yield spread accounts for almost the entire yield differ-ential (and its variability) from 2009 to mid-2011. Then, in the midst of theEuropean debt crisis, a non-default component of about 40 basis pointsemerges; however, this component bears no close relationship to the“euro risk” factor. By contrast, Finland and the Netherlands exhibit asmall, volatile and at times negative CDS differential relative toGermany, with the non-default component accounting for the largestpart of the yield differential. It is however difficult to interpret this

–.01

.00

.01

.02

.03

.04

.05

.06

.07

2008 2009 2010 2011 2012 2013 2014 2015

Yields Componenents ES

–.01

.00

.01

.02

.03

.04

2008 2009 2010 2011 2012 2013 2014 2015

Yields Componenents BG

–.02

.00

.02

.04

.06

.08

.10

.12

2008 2009 2010 2011 2012 2013 2014 2015

yield spread vs GERcds spread vs GERnon default component vs GERredenomination risk

Yields Componenents IR

–.01

.00

.01

.02

.03

.04

.05

.06

2008 2009 2010 2011 2012 2013 2014 2015

Yields Componenents IT

–.02.00.02.04.06.08.10.12.14.16

2008 2009 2010 2011 2012 2013 2014 2015

Yields Componenents PT

Figure 3b: The default and non-default component in yields spreads – Low Yielders.Source: Datastream/Thomson Financial.




non-default component as representing a liquidity premium as it corre-lates with the “euro risk” factor, at least until the beginning of 2012.

iv. Until 2011 the non-default component of the yield spread is small for allmember States, not exceeding 50 basis points (except in Finland), but itbecomes sizeable in the midst of the European debt crisis, fluctuatingaround or above 100 basis points in the case of high yielders. Thesimultaneous increase in the “euro risk” factor suggests that euro deva-luation or break-up risk, rather than liquidity risk, is the likely explana-tion of such large non-default components.

v. For all countries, before 2010 and after 2013, non-default components areunlikely to reflect the risk of devaluation or a break-up of the euro, and thusprovide a rough estimate of liquidity premia. During the global financialcrisis the non-default component is sizeable in Finland, reaching 70 basispoint, and not negligible, around 40–50 basis points, in Belgium, theNetherlands, and France. However, in the aftermath of the European debtcrisis, the component of the spread likely due to liquidity problems becomesvery small, around 25 basis points in Finland and the Netherlands, and evenlower, around 10 basis points, in Austria and France.

Our analysis suggests that the greater liquidity of Eurobonds (if introduced on alarge scale) would mostly benefit small issuers with some reduction in theirborrowing costs. However, as we have just looked at interest rate differentialsrelative to Germany, we cannot exclude that liquidity might improve for Germanyand the other large issuers as well, thus determining a generalized reduction inthe level of interest rates. On the other hand, we are skeptical that any suchreduction would be significant for at least two reasons. First, it is hard to imaginea further reduction in interest rates below their currently negative levels. Second,as shown by Favero and Missale (2012), the liquidity premium of about 40 basispoints that German Bunds historically paid on US Treasuries, has reversed in favorof Germany after the global financial crisis. To conclude, reductions in borrowingcosts from a potentially larger and more liquid market for Eurobonds are likely tobe small and mostly appealing to small issuers. Arguments for Eurobonds basedon liquidity enhancements and, consequently, lower costs appear weak, althoughthey cannot be completely dismissed.

4 Fiscal fundamentals, global risk and contagion

Having shown that default risk is the main driver of yield spreads with someadditional role for euro devaluation and break-up risk, in this section we




examine the relative importance of fiscal fundamentals, global risk factors andcontagion. The analysis follows Favero and Missale (2012) where we findstrong evidence that changes in market sentiment significantly affect yieldspreads in the Euro area and have an important role in propagating thedebt crisis.

We focus on the default premium component of yield spreads, and model itas a linear function of local fiscal fundamentals, i. e. the expected debt-to-GDPand deficit-to-GDP ratios, the “euro risk” factor, and a global risk factor that wecall the “global spread”, which interacts the spreads of the other member Stateswith local fiscal fundamentals to capture the global risk that a country faces.The global spread variable is country-specific and is obtained as the weightedaverage of other countries’ spreads with weights reflecting fiscal proximity, thatis, the distance of their debt and deficit ratios to the same variables of thecountry considered; the closer the debt and the deficit of the country consideredto those of another country, the greater the weight assigned to the yield spreadof that country. This makes the exposure of each country to the spreads of theother countries in the euro area depend on the “distance” between their fiscalfundamentals. The main novelty relative to Favero and Missale (2012) is theintroduction of a common “euro risk” factor, i. e. the average Quanto CDS, toconsider the risk of a euro break-up.6 Finally, we split our sample into two sub-periods, before and after August 2012, in order to assess the impact on theestimated coefficients of the ECB announcement on August 2 of OMTs, i. e.secondary market purchases of government bonds of member States receivingfinancial assistance from the ESM.

4.1 A GVAR model of yield spreads

We estimate by Seemingly Unrelated Regressions (SUR) the following 10-equa-tions (countries) Global VAR (GVAR) for the 10-year yield spreads on GermanBunds for Austria, Belgium, Finland, France, Greece, Ireland, Italy, theNetherlands, Portugal and Spain, using daily data over the period November2009 to November 2015:

6 We restrict our analysis to the euro debt crisis period and use daily data, while Favero andMissale (2012) used weekly data over the period June 2006–August 2011. As a result, we do notinclude in our specification global factors such as the US Baa-Aaa bond spread, which is verystable over the restricted sample period.




Yit −Y

Gt

� �= β0 + β1 Yi

t − 1 −YGt − 1

� �+ β2 bit − b

Gt

� �+ β3 dit −d

Gt

� �+

+ β4ERt + β5ERt − 1 + β6GSit + β7GS

it − 1 + u

it

[1]

GSit =Xj≠ i

wijt Yj

t −YGt

� �

distijt =0.5 bit − bjt

� �=60+ 0.5 dit −d

jt

� �=3

qijt =1

distijtif distijt�� < 1;0 otherwise

wijt = q

ijt

Pj≠ i q

ijt

.The model relates yield spreads, i. e. the difference between the yield Yi

t of acountry i and the German yield YG

t , to the difference in debt- and deficit-to-GDPratios, bit − b

Gt and dit −d

Gt , the common ‘euro risk’ factor, ERt, and the global

spread GSit.Following Attinasi et al. (2010), for fiscal fundamentals we consider the

average for a 2-year period of the expected deficit-to-GDP ratio, dit, and theexpected debt-to-GDP ratio, bit. The expected variables are the EuropeanCommission Forecasts that are released on a bi-annual basis. We include inthe model the difference between each country’s forecast and the forecast of thesame variables for Germany. The common ‘euro risk’ factor, ERt, that capturesthe euro devaluation risk and possibly the risk of a euro break-up, is obtained asthe simple average of the Quanto CDS of the countries under consideration, asexplained in Section 3.2. The global spread, GSit, captures the global (euro area)risk that a country faces. It is constructed as a country-specific stochastic trendthat is mostly driven by the yield spreads of the countries that have fiscalfundamentals more similar (or less distant) to the country considered.Specifically, the global spread is measured for each country by a weightedaverage of yield spreads of other countries. Weights, wij

t , are the inverse of thefiscal distance, distijt , which is defined as the absolute value of the differencebetween the fiscal fundamentals, dt and bt, of two countries, i and j (normalizedby their 3 and 60 percent limits). Hence, weights are constructed to make theglobal spread more dependent on the spreads of those countries that are moresimilar in terms of fiscal fundamentals. The time-varying weights, related to thechanging forecasts for fiscal fundamentals, have the potential of explaining thechanging correlation of spreads observed in the descriptive analysis of theprevious section. The global spread variable is inspired by the construction ofglobal variables in the GVAR modeling approach (see, e.g., Pesaran et al., 2004;Dees et al., 2007), where global macro variables are constructed for each country




by using trade weights; we simply replace trade shares with the above measureof fiscal proximity, wij

t . Using the distance in terms of fiscal fundamentals makesthe global spread country-specific and the weights not constant, unlike instandard GVAR based on trade shares. Note that the global spread variablesvary at daily frequency only for the fluctuations in the yield spreads, while thelow frequency fluctuations depend both on the fluctuations in the yield spreadsand on the changes in the weights. As the latter occurs only biannually, noendogeneity problem for the weights should arise.

We estimate the model over the full sample and over two subsamples: fromNovember 2009 to 1 August 2012 and from 2 August 2012 to November 2015. Thesplit in subsamples is introduced to assess the impact of the ECB announcementof OMTs on the relation between yield spreads and their determinants; localfiscal fundamentals, global spreads, and the “euro risk” factor.

4.2 The empirical evidence

The results of the estimation, which are reported in Table 1 in the Appendix, canbe summarized as follows:

i. All spreads are very persistent. This implies that long-run means are veryimprecisely estimated because yield spreads tend to display a stochastictrend that first slopes upward and then downward.

ii. The effect of fiscal fundamentals, when they are not interacted with theyield spreads of other member States, is rarely significant. The debt ratiois significant only in Ireland over the second period, and in the case oflow yielders but with the wrong sign. The deficit is significant only inIreland and Portugal over the first sample period, and in Austria andSpain over the second period.

iii. The common “euro risk” factor (capturing the risk of a euro break-up) isstrongly significant for all countries, except for Portugal, before the ECBannouncements of OMTs while its decline over the second period signifi-cantly reduces the yield spreads of high yielders. Its impact on the yieldspread of the high yielders is also stronger and more persistent than onthe spread of low yielders.

iv. The global spread variable that makes the exposure to other countries’yield spreads increase with the similarity of their debt and deficit ratios(so as to link more closely the spreads of countries with similarly weakfundamentals) is strongly significant in all member States with differentimpact coefficients. The impact of the global spread is stronger on theyield spread of high yielders than on the spread of low yielders.




v. The ECB announcement of OMTs is associated with a steady decline ofyield spreads and the common “euro risk” factor, and with a change intheir relationship that differs across countries. In the case of low yielders,the impact of the “euro risk” factor becomes either negative or notsignificant whereas it remains significant and gains strength in the caseof high yielders.

vi. Panel restrictions cannot be imposed on the system as the coefficientsdiffer importantly across countries.

The evidence in Table 1 shows that global spreads are highly significant forall countries considered while fiscal fundamentals have no effect at any con-ventional significance level, with the exception of Ireland and Portugal and inonly one of the two subsamples. Fiscal fundamentals matter not per se butbecause they determine the exposure of each country’s yield spread to theyield spreads of the other member States. Fiscal proximity selects the referencegroup of countries whose yield spreads determine the global spread to which acountry is exposed. Member States with sound fiscal fundamentals areimmune to the risk priced in the yield spreads of countries with fiscal problemswhile member States with weak fundamentals are affected by the yield spreadsof troubled countries to the extent that they are fiscally similar. Hence, fiscalfundamentals matter in the pricing of default risk but only as they determinethe exposure to other countries’ spreads, that is, to the global risk that themarket perceives. When global risk factors are muted, fiscal fundamentalshave no effect on yield spreads. This evidence suggests that markets do setincentives for fiscal discipline but they do it discontinuously, namely onlywhen global systemic risk is perceived.

The effect of the ECB announcement of OMTs is found by comparing theresults of the estimation of the GVAR model over the two subsamples pre andpost August 2012. To this end it is worth noting that the period following the ECBannouncement is characterized by a steady decline in the “euro risk” factor, asthe risk of a euro break-up fell. This downward trend actually starts one weekbefore, on July 26, following Mario Draghi’s statement that he would do “what-ever it takes” to defend the euro. Furthermore, as shown in Figure 3(a) and (b),the announcement of OMTs on 2 August 2012 marks the turning points in theyield spreads of Italy and Spain, the two countries under greater debt-financingstress at that time. In other member States, borrowing conditions seem insteadto have eased earlier, in the first half of 2012.

The ECB announcement of OMTs clearly led to a change in the relationshipbetween the yield spreads and the risk of a euro break-up that differs acrosscountries. In the case of low yielders, the impact of the “euro risk” factor




becomes either negative or not significant whereas it remains significant andgains strength in the case of high yielders. We interpret the result for the formercountries as evidence that the ECB announcement does not only reduce eurodevaluation risk and the probability of a break-up but it also makes the yieldspread of countries with sound fiscal fundamentals immune to such risks. Bycontrast, sensitivity to the “euro risk” factor increases in countries with weakerfiscal fundamentals. As fundamentals are expected to change slowly over time,a lower devaluation risk is the main factor that drives their yield spreads down.In fact, after the ECB announcement, the yield spreads of high yielders withcloser fiscal fundamentals decrease at the same pace, leaving the estimatedcoefficient on the global spread statistically unaffected with the notable excep-tion of Spain, where the yield spread displays some resilience, reflecting theproblems of its troubled banking sector. This evidence contrasts with that forlow yielders, where the estimated sensitivity to the global spread increasessignificantly after the ECB announcement of OMTs. As the risk of a euro break-up vanishes, it appears that the yield spreads of countries with sounder fiscalfundamentals tend to co-move more synchronously with each other and withthose of weaker member States.

Summing up, there is strong evidence of country interdependence in thepricing of default risk, which suggests that a crisis can easily propagate fromcountries with weak fiscal fundamentals to fiscally sounder member States.However, the channel of yield interdependence differs between high yieldersand low yielders and over the two sub-periods. The former countries are linkedthrough global spreads; i. e. they are exposed to the yield spreads of othertroubled countries to a degree which increases with fiscal proximity.7 Low yielderswith sounder fiscal fundamentals are partially immune to the high spreads offiscally weak countries but are still exposed to the risk of a devaluation or break-up of the euro, as priced in Quanto CDS. This “euro risk” factor is the main driverof the yield spreads of low yielders until August 2012, which suggests that abroader crisis of the euro area is the channel through which a debt default in acountry with weak fundamentals can propagate to safer member States.

This evidence suggests that the introduction of Eurobonds would have beenin the interest not only of countries with debt-financing difficulties but also ofcountries with stronger fundamentals in that a debt default of, say, Italy orSpain would rapidly spread across the whole Monetary Union. However, forEurobonds to be justified as an instrument for crisis prevention, two conditionsmust be satisfied. The first one concerns the nature of the debt crisis; only in thecase of a self-fulfilling liquidity crisis not due to fundamentals, Eurobonds

7 High yielders are also affected by the “euro-risk” over the entire sample period.




would be economically justified and mostly effective at stabilizing investors’expectations, thus avoiding a debt run. Although fiscal fundamentals do notaffect yield spreads directly, we have shown that they determine the exposure toother countries’ yield spreads, that is, to the global risk that the market per-ceives. The issue to which we turn in the next section is whether the impact ofthe global spread variable is constant or dominated by changes in marketsentiment. The second condition for Eurobonds is the absence of a “lender oflast resort” who might solve a liquidity crisis. With the announcement of OMTs,the ECB seems to have taken such role upon itself, mainly as a response to theincrease in the risk of a euro break-up. The strong reaction of the market,leading to a dramatic fall in Quanto CDSs and yield spreads, is itself evidenceof the non-fundamental nature of the crisis. On the other hand, the ESM–ECBplan for crisis prevention has most likely eliminated the need for Eurobondseven if the ECB cannot act as a genuine lender of last resort because a necessarycondition for OMTs is that a country receives financial assistance from the ESMand has thus committed to an adjustment program with policy conditionsattached.8 We turn to this issue in the last section.

4.3 Contagion in the EMU

Eurobonds are justified as instruments for crisis prevention if market sentimentdominates the pricing of default risk. In fact, evidence of significant changes inmarket sentiment would provide a good indicator of the likelihood of a debtrun, i. e. of a liquidity crisis driven by self-fulfilling expectations. Thus, therelevant issue is whether the market’s assessment of sovereign risk in relationto fiscal fundamentals and global risk is constant, and thus reliable, or subjectto shifts in sentiment; in other words, whether the impact of the global spreadvariable is stable or changing over time. To address this issue we look at thestructural stability of the coefficient on the global spread variable. This iscrucial to assess the presence of contagion. Indeed, time variation in theimpact of the global spread variable on domestic yield spreads would implythat shifts in market sentiment dominate the fundamentals-driven interdepen-dence across countries.

8 Moreover, the ECB cannot purchase government bonds in the primary market, i. e. atissuance, as Art. 123(1) TFEU prohibits the ECB from providing any type of credit facility togovernments or other public authorities, and from purchasing debt instruments from themdirectly (see footnote 1 for further details).




To measure the effect of contagion we study the case of Italy, estimating amultivariate GARCH model of two equations for the yield spread and the asso-ciated global spread. The estimated reduced form specification of our GlobalVAR is

YITt −YG

t

� �GSITt

" #=B0 +B1

YITt − 1 −Y

Gt − 1

� �GSITt − 1

" #+B2

FFITt − 1

GFFt − 1

" #+B3ERt − 1 +H

1=2t

εITtεGSt

" #

[2]

vech Htð Þ=M +Avech εt − 1ε′t − 1� �

+Bvech Ht − 1ð ÞThis specification models the joint process of the Italian yield spread and theglobal spread variable relevant to Italy, GSITt , as a persistent process with a meandetermined by the expected Italian fiscal fundamentals FFIT

t − 1. The latter are thesame debt and deficit ratios relative to Germany used in system specification (1),while GFFt − 1 is the weighted average of the fiscal fundamentals relative toGermany of other countries with weights determined by fiscal proximity toItaly as in the global spread variable.

The identification of the structural parameters is achieved by a triangular-ization of the variance-covariance matrix of the residuals, Ht, based on theassumption that the Italian yield spread is contemporaneously caused by theglobal spread but not vice-versa. The model in equation (2) allows for a time-varying conditional variance-covariance between the Italian yield spread andthe corresponding global spread in that the Ht matrix is updated every period.In particular, the time-varying variance-covariance matrix of the residuals, Ht, ismodelled as a diagonal BEKK (Engle and Kroner, 1995) system. This specificationcan, then, be used to generate a time-varying estimate of the impact of theglobal spread on the Italian yield spread.

The model provides us with a natural measure of contagion: the dynamicconditional beta in the terminology of Bali and Engle (2010), which is thecoefficient γt determining the effect of a shock in the global spread on theItalian spread:

E εITt jεGSt� �

= γtεGSt withγt = h12, th

− 122, t

where the hij, t are the estimated elements of the Ht matrix.Variations in the coefficient γt are reported in Figure 4 along with the

estimates of the global spread coefficient obtained by SUR estimation of theGVAR model in equation (1), and displayed as a continuous line for the twosubsamples, November 2009–July 2012 and August 2012–November 2015.

The impact of the global spread on the Italian yield spread varies signifi-cantly over time, a clear sign of changes in market sentiment. During the




European sovereign debt crisis, the sensitivity of the Italian yield spread to itsglobal spread became much higher than it would have been in the case of aconstant market reaction to global risk, as implied by SUR estimation of theGVAR model.

Figure 5 provides an estimate of the effect of contagion as measured by thedifference between the impact effect of the global spread as estimated in themultivariate GARCH model and that obtained from the constant parameter SUR

–0.4

0.0

0.4

0.8

1.2

1.6

2.0

2009 2010 2011 2012 2013 2014 2015

Figure 5: The impact effect of contagion on the yield spread Italy-Germany.Sources: Authors’ calculation.

–0.4

0.0

0.4

0.8

1.2

1.6

2.0

2.4

2009 2010 2011 2012 2013 2014 2015

impact of a global trend shock on the Yield spread Italy-Germany (MGARCH)impact of a global trend shock on the Yield spread Italy-Germany (SURE)

Figure 4: Interdependence and contagion between the yield spread Italy-Germany and theglobal spread.Sources: Authors’ calculation.




estimation. Specifically, in Figure 5, we report the difference of the responses (ofthe yield spread) in the two models to a given increase in the global spread of100 basis points.

In the second half of 2011, the effect of contagion became sizeable.Figure 5 shows that, because of market sentiment, the estimated impacteffect on the Italian yield spread is almost 200 basis points higher. Thisevidence suggests that in late 2011, the Italian yield spread started to deviatesignificantly from its fundamentals-driven equilibrium and its fluctuationsbecame increasingly dominated by shifts in market sentiment. Favero andMissale (2012) report similar evidence for Spain. Financial markets not onlyexert their disciplinary role discontinuously but their overreaction to globalrisk is itself an important source of instability. In late 2011 and early 2012,market sentiment played a major role in the pricing of default risk in Italyand Spain, a fact which called for “a lender of last resort” to halt the debtrun and its spreading to safer member States.

4.4 The ECB announcement of OMTs

On 2 August 2012, the ECB announced the possibility of OMTs, that is, unlimitedsecondary market purchases of bonds of member States receiving financialassistance from the ESM under a program that includes ESM primary marketpurchases.

As shown by the GVAR estimation over the two subsamples and discussedin Section 4.2, the ECB announcement of OMTs led to a steady decline inQuanto CDSs and to a significant fall in the yield spreads of Italy and Spain.The strong influence that the ECB announcement had on market expectationsis well evidenced by our estimate of the effect of contagion based on thedifference between the impact effect of the global spread as estimated in themultivariate GARCH model and in the constant parameter SUR system. Figure 5shows that the impact effect associated with shifts in market sentiment issignificantly lower in the second subsample, pointing to a substantial effecton market expectations of the ECB’s willingness to act as a quasi-lender of lastresort. We interpret this finding as further evidence of the non-fundamentalnature of the Italian debt crisis in 2011–2012 and the role of self-fullingexpectations in making a run equilibrium possible. Quite revealingly of theself-fulfilling nature of the Italian crisis, no bonds have yet been purchasedunder the OMT program.

The strong effect that the OMT announcement had on investors’ expecta-tions and interest rates shows the importance of a lender of last resort in




insuring the market against the risk of a liquidity crisis. However, to the extentthat the ECB’s ability to buy member States’ bonds reduces the risk of a self-fulfilling debt run, it also deprives Eurobonds of a role in crisis prevention. Theimplications of the ESM–ECB scheme for Eurobonds are discussed in the nextsection.

5 The crisis-prevention role of Eurobondswith the ESM–ECB scheme

A lender of last resort, prepared to buy unlimited amounts of a member State’sbonds if necessary, is the best way to halt a debt run and prevent a propagationof the crisis. If the central bank cannot act as a lender of last resort, thepossibility to issue Eurobonds in case of a debt run would ensure market accessto a country with financing difficulties and halt a debt crisis that would rapidlypropagate to other countries. In particular, the yield on Eurobonds would beimmune to the changes in market sentiment observed in our econometricanalysis. Eurobonds jointly guaranteed by all member States would ensure thefunding needed to roll over the maturing debt at a reasonable cost and givetime, and financial breathing space, for fiscal adjustment. To avoid moralhazard, the right to issue Eurobonds could be granted only to member Statessatisfying ex-ante conditions regarding their debt sustainability. In order toprevent countries with high interest rates from issuing Eurobonds to reduceborrowing costs at the expense of relatively safer States, their use could belimited to crisis situations, say, when the yield spread observed in the secondarymarket exceeds a pre-specified threshold. Then, the mere right to issueEurobonds, if credible, would be enough to avoid a debt run, i. e. a panicequilibrium where each investor refuses to roll over the debt for fear that any-body else would do the same. If the crisis were one of liquidity, no Eurobondwould have to be issued.

The establishment of the ESM, and the design of the ESM–ECB scheme forfinancial assistance, have probably made Eurobonds unnecessary as an instru-ment for crisis prevention. In fact, a country facing a debt run can now rely onthe ESM Primary Market Support Facility, i. e. the drawdown of an ESMprecautionary credit line for bond purchases on the primary market (up to50% of the issue amount), and on OMTs, i. e. the ECB’s unlimited purchase ofbonds with a maturity between 1 and 3 years on the secondary market. Ifcredible, the joint intervention of the ESM in the primary market and the ECB




commitment to buy bonds in the secondary market removes the risk of a self-fulfilling debt run.

The most compelling evidence of the insurance that the ECB can provideagainst a self-fulfilling debt run is the market reaction to the ECB GoverningCouncil’s announcement on 2 August 2012 that the ECB would undertakeOMTs in the secondary bond market.9 The announcement gave content toMario Draghi’s statement that he would do “whatever it takes” to defend theeuro, and triggered a rapid decline in yield spreads of Italian and Spanishbonds that fell from about 500–300 basis points in less than three months.This dramatic reduction is well captured by the increased sensitivity of yieldspreads to the “euro risk” factor and by the effect of improved marketsentiment shown in Figure 5. This evidence supports the view of DeGrauwe (2011, 2013) that the lack of a lender of last resort was what themarket feared the most. Importantly, the announcement of OMTs has so farworked by stabilizing investors’ expectations with no need for bond pur-chases by the ECB.

The strong effect that the OMT announcement had on investors’ expecta-tions and interest rates shows the reassurance provided by the Central Bank’ability to buy debt. However, if a lender of last resort is what the marketwants to solve the coordination problem and avoid a debt run, it is worthnoting that the OMT program does not make the ECB a genuine lender of lastresort. For the ECB to intervene with bond purchases on the secondarymarket, a member States that needs financial assistance must first activatean ESM precautionary program, i. e. a credit line to be used for ESM bondpurchases in the primary market under a Primary Market Support Facility.Precautionary assistance from the ESM is subject to the fulfillment of elig-ibility criteria, i. e. ex-ante conditionality, and the signing of a memorandumof understanding, i. e. ex-post conditionality. Eligibility criteria include asustainable public debt, a sustainable external position and a sound finan-cial system. Ex-post conditionality refers to specific policy actions, e. g. fiscaladjustment and reforms, to achieve the objectives indicated in the memor-andum of understanding.

Introducing ex-post conditionality (in addition to eligibility criteria) wasnecessary to overcome political resistance against ECB intervention, but it hasnegative consequences. First, conditioning ECB intervention on fiscal and policyadjustment appears unnecessary when a country already satisfies the eligibility

9 The technical framework of these operations was formulated on 6 September 2012.




requirements of sustainable public debt and a sound financial system. If the ex-ante condition is met, and the crisis is one of liquidity, the requirement of fiscaladjustment can send the wrong signal and impair the ability of the ECB to act asa lender of last resort when this is economically justified. Secondly, and moreimportantly, because of the stigma implied by the imposition of fiscal measures,a country can be reluctant to ask for assistance even in the midst of a liquiditycrisis when timely intervention by the Central Bank is crucial (see, e. g. VandenBosh 2012).10

Clearly, if the ESM-ECB framework were ever tested and proved to be toorestrictive to provide rapid and effective liquidity assistance either because ofpolitical delays in the request for assistance or in the design of the adjustmentprogram, then the case for Eurobonds to maintain market access in crisissituations would strongly re-emerge. However, at present, this appears a remotepossibility.

6 The role of Eurobonds – concluding remarks

Despite its rigidity, the ESM–ECB scheme for precautionary financial assistancehas so far been successful in stabilizing investors’ expectations and halting therun on the Italian and Spanish debt. In any case, constraints preventing the ECBfrom acting rapidly in crisis situations should not be an argument for introdu-cing Eurobonds, but rather an incentive for a reform that makes ESM financialassistance available on the fulfillment of ex-ante conditionality, as in the case ofIMF assistance under the Flexible Credit Line.

More importantly, the introduction of Eurobonds as a crisis instrumentfaces strong political opposition because it raises moral hazard issues, andEurobonds have economic costs that safer member States are not willing topay. Giving a member State the option to issue a bond jointly guaranteed byall other States, even if not exercised, is an implicit contingent liability forStates with stronger fiscal fundamentals. While such costs are certain andimmediate, the gains from increased liquidity will eventually arise only in adistant future, as they require that a market for Eurobonds as large and deepas that of US Treasuries develops. It is then not surprising that Eurobondproposals have faced the strong opposition of safer member States like

10 Other features of the ESM–ECB scheme, such as the limitation of OMTs to bonds in thematurity segment between 1 and 3 years, appear unduly restrictive and may limit the scope ofECB intervention.




Finland, Germany and the Netherlands. As the benefits for the latter coun-tries would arise mainly because of a stronger Monetary Union, Eurobondswill never be used for crisis prevention without the aim for greater fiscalunity. The solidarity, cooperation and political foresight that would beneeded for jointly guaranteed EU debt are simply not present in the EU atthe moment.

If the opposition to the use of Eurobonds for deficit financing and debt roll-over is understandable, the hope for jointly guaranteed EU debt can stay alive ifEurobonds are linked to specific projects or expenditures that provide benefits toall member States. A small scale program for financing investments or expendi-tures that have positive externalities and spillover effects could gain politicalconsensus and have some chance of success.

The idea is to issue Eurobonds in order to finance projects and/or expendi-tures that provide benefits to all member States also because of spillover effectsand/or externalities. This is clearly the case for EU infrastructure projects, but asimilar argument can be made for any public investment that increases outputacross the EU through spillover effects. More recently, it has been argued thatextraordinary expenditures to deal with the refugee crisis and/or to enhancesecurity should be financed with Eurobonds. As these problems are common toall member States and the expenditures incurred by each member State haveclear positive externalities to other States, Eurobonds seem the appropriateinstruments to share the borrowing cost and credit risk associated with theirfinancing.

Furthermore, issuance of Eurobonds backed by several and joint guaran-tees would represent an important first experiment to test investors’ demandand borrowing costs. A common Eurobond program would enhance thecredibility of EU institutions by signaling a political will for greater fiscalunity and cooperation, thus paving the way for a deeper reform of EU fiscalgovernance.

Acknowledgements: We thank participants at the Workshop on the Law andEconomics of “Eurobonds Beyond Crisis Management” (Hamburg, 8–9 January2016) for useful comments and suggestions. Helpful comments from two anon-ymous referees are gratefully acknowledged.




App

endix

Table1:

Yieldsp

read

son

bund

s,seem

inglyun

relatedregression

–su

bsam

pleeviden

ce,da

ilyda

ta.HighYielde

rs:Spa

in,Greece,

Irelan

d,Italy,

Portug

al.

ESP

GRE

IRL

ITA

POR

Sam

ple

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

Con

stan

t.E

–

E

.E

.

.

.E

.

.

–.E

.E

.

.

.E

.

.E

(E

)

(.)

(.E)

(.)

(.)

(.)

(.)

(.)

(:E)

(.E)

(.)

(.)

(.E)

(.)

(.)

Yi t−

1−YG t−

1.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

bi t−bG t

–.E

.

–E

–

.E

–.

–.

–.E

.

.

–.E

–.

–.

–.E

.

–.E

(.)

(.)

(:E)

(.)

(.)

(.)

(.E)

(.)

(.)

(.E)

(. )

(.)

(.E)

(.)

(.)

di t−dG t

.

–.

.

.

–.

–.

.

.

.

–.

.

–.

.

.

–.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

( .)

(.)

(.)

ERt

.

.

.

.

.

–.

.

.

.

.

.

.

.

–.

.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

ERt−

1–

.

–.

–.

–.

–.

.

–.

–.

–.

–.

–.

–.

–.

–.

–.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

GSi t

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

GSi t−

1–

.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

( .)

(.)

(.)

(.)

(.)

Adj

R-sq

uared

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Mea

nDep

.Variable

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

SEof

Regression

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(con

tinu

ed)




Table1:

(con

tinu

ed)

BEL

FIN

FRA

NL

OE

Sam

ple

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

:

Con

stan

t.E

.E

.E

.E

–

.E

.E

.E

–.E

.E

.E

–.E

.E

.E

–.E

.E

(.E

)

(.E

)

(.E

)

(.E

)

(.E)

(.E

)

(.E)

(.E)

(.E)

(.E)

(.E)

(.E)

(.E)

(.E)

(.E)

Yi t−

1−YG t−

1.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

bi t−bG t

–.E

–.

.

–.E

–.

–.E

.E

–.E

–.E

–.E

–.E

–.E

–.E

–.

.E

(.)

(.)

(.)

(.E)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

di t−dG t

–.E

–.

.

–E

.E

–.E

.E

–.E

.

–.E

–.E

.

.E

–.

.

(.)

(.)

(.)

(.)

(.E)

(.E)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

ERt

.

.

.

.

.

–.

.

.

.

.

.

–.

.

.

.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(. )

(.)

ERt−

1–

.

–.

–.

–.

–.

.

–.

–.

–.

–.

–.

.

–.

–.

–.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

GSi t

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

GSi t−

1–

.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

–.

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

(.)

( .)

(.)

(.)

(.)

(.)

Adj

R-sq

uared

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Mea

nDep

.Variable

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

SEof Regression

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Note:

Coe

fficientssign

ifican

tat

the5pe

rcen

tlevelus

ingthestan

dard

t-distribu

tion

arerepo

rted

inbo

ld.

Yi t−YG t

�� =

β 0+β 1

Yi t−

1−YG t−

1

�� +

β 2bi t

−bG t

�� +

β 3di t

−dG t

�� +

β 4ER

t+β 5ER

t−1+β 6GSi t+β 7GSi t−

1+ui t




References

Attinasi, M.G., C. Checherita, and C. Nickel. 2010. “What Explains the Surge in Euro AreaSovereign Spreads During the Financial Crisis 2007–2009?,” 10 Public Finance andManagement 595–645.

Bali, T.G., and R.F. Engle. 2010. “Resurrecting the Conditional CAPM with Dynamic ConditionalCorrelations,” SSRN Working Paper, http://papers.ssrn.com/sol3/papers.cfm?abstract_id= 1298633.

Bernoth, K., J. von Hagen, and L. Schuknecht. 2012. “Sovereign Risk Premiums in the EuroGovernment Bond Market,” 31 Journal of International Money and Finance 975–995.

Calvori, F., M. Dentella, and G.P. Gallo. 2016. “Sovereign Debt Spreads in the Euro Area:When Fears Become Excess Fears,” Working Paper No 3/2016, DISIA Università di Firenze.http://local.disia.unifi.it/wp_disia/2016/wp_disia_2016_03.pdf.

Dees, S., F. Di Mauro, M.H. Pesaran, and L.V. Smith. 2007. “Exploring the International Linkagesof the Euro Area: A Global VAR Analysis,” 22 Journal of Applied Econometrics 1–38.

De Grauwe, P. 2011. “Managing a Fragile Eurozone,” 12 CESifo Forum 40–45.De Grauwe, P. 2013. “The European Central Bank as Lender of Last Resort in the Government

Bond Markets,” 59 CESifo Economic Studies 520–535.De Santis, R.A. 2015. “A Measure of Redenomination Risk,” Working Paper Series No. 1785,

European Central Bank. https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1785.en.pdf.Di Cesare, A., G. Grande, M. Manna, and M. Taboga. 2012. “Recent Estimates of Sovereign Risk

Premia for Euro-Area Countries,” Occasional Papers No. 128, Bank of Italy. https://www.bancaditalia.it/pubblicazioni/qef/2012-0128/QEF_128.pdf.

ECB. 2012. “Technical Features of Outright Monetary Transactions,” Press Release, September6, http://www.ecb.europa.eu/press/pr/date/2012/html/pr120906_1.en.html.

Engle, R.F., and K.F. Kroner. 1995. “Multivariate Simultaneous Generalized ARCH,” 11Econometric Theory 122–150.

Favero, C.A. 2013. “Modelling and Forecasting Government Bond Spreads in the Euro Area:A GVAR Model,” 177 Journal of Econometrics 343–356.

Favero, C.A., and A. Missale. 2010. “EU Public Debt Management and Eurobonds,” in EuropeanParliament’s Committee on Economic and Monetary Affairs, ed. Euro Area Governance –Ideas for Crisis Management Reforms. Brussels: European Parliament.

Favero, C.A., and A. Missale. 2012. “Sovereign Spreads in the Euro Area. Which Prospects for aEurobond?,” 27 Economic Policy 231–273.

Giovannini Group Report. 2000. “Co-ordinated Public Debt Issuance in the Euro Area.”November. http://ec.europa.eu/economy_finance/publications/publication6372_en.pdf.

Monti, M. 2010. “A New Strategy for the Single Market. At the Service of Europe’s Economy andSociety,” Report to the President of the European Commission. http://ec.europa.eu/bepa/pdf/monti_report_final_10_05_2010_en.pdf.

Pesaran, M.H., T. Schuermann, and S.M. Weiner. 2004. “Modelling Regional InterdependenciesUsing a Global Error-Correcting Macro-Econometric Model,” 22(2) Journal of Business andEconomic Statistics 129–162.

Vanden Bosch, X. 2012. “Preventing the Rise of Sovereign Borrowing Costs in the Eurozone:What Can the ESM and the ECB Achieve?” Egmont Paper No. 56, Royal Institute forInternational Relations. http://aei.pitt.edu/39375.




carlo a. favero* and alessandro missale contagion in the ... · carlo a. favero* and alessandro...

Documents